Flow Measurement and Instrumentation 22 (2011) 215–224

Contents lists available at ScienceDirect

Flow Measurement and Instrumentation

journal homepage: www.elsevier.com/locate/flowmeasinst

Flow pattern and hydraulic performance of the REDAC Gross Pollutant TrapAminuddin Ab Ghani a, H.Md. Azamathulla a,∗, Tze Liang Lau b, C.H. Ravikanth a, Nor Azazi Zakaria a,Cheng Siang Leow a, Mohd Azlan Mohd Yusof aa River Engineering and Urban Drainage Research Centre (REDAC), Universiti Sains Malaysia, Engineering Campus, Seri Ampangan, 14300 Nibong Tebal, Penang, Malaysiab School of Civil Engineering, Universiti Sains Malaysia, Engineering Campus, Seri Ampangan, 14300 Nibong Tebal, Penang, Malaysia

a r t i c l e i n f o

Article history:Received 27 April 2010Received in revised form22 February 2011Accepted 28 February 2011

Keywords:Gross Pollutant TrapSedimentBest management practiceFlow measurement

a b s t r a c t

This paper discusses the flow pattern and hydraulic performance of a Gross Pollutant Trap (GPT),designed and patented by River Engineering and Drainage Research Centre (REDAC) at Universiti SainsMalaysia. Stormwater problems have become more severe due to the increase in urbanization. Theincrease in the amount of impervious surface in urban areas produces more stormwater runoff, that iscarried to the receiving bodies of water. The higher runoff volume also carries more pollutants (grosspollutants, sediments, andnutrients) from the contributing catchment area. Coarse sediments transportedby stormwater runoff have negative effects on the receiving body of water and the aquatic environmentby covering up aquatic habitats and clogging waterways. One of the challenges in designing a GPT forurban stormwater drainage is providing effective trapping without hindering the hydraulic functionof the channel, thus, avoiding overspill or flooding. The current study presents a GPT design to meetthese specific requirements of trapping efficiency and hydraulic function. The current GPT overcame thecommon problem of overspilling of gross pollutants in GPT by the introduction of additional overspillcompartments that can handle excessive runoff and improve pollutant trapping in higher flow conditions.In laboratory testing, the prototype GPT was capable of achieving good trapping efficiency (over 80% forgross pollutants and over 60% for coarse sediments) without causing any overspill.

© 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Gross pollutants are defined as discarded materials largerthan 5 mm and include litter and debris, and coarse sedimentsare particles with grain sizes greater than 0.5 mm [1]. Litter inthe stormwater system includes human-derived rubbish, suchas paper, plastic, polystyrene, metal, and glass, that has beendumped,mainly illegally, intowaterways or drains. Debris consistsof organic materials including leaves, branches, twigs and grassclippings. Generally, coarse sediments are eroded soil particlesoriginating from diffused sources, such as construction sites,land clearing sites and agricultural areas. All of these pollutantscan potentially harm wildlife, especially those found in aquatichabitats, and can decrease the aesthetic qualities of the stormwatersystem and attract vermin. Fig. 1 shows the various types of grosspollutants trapped in receiving bodies. Based on the photographs,the gross pollutants were classified into several types of bottles,

∗ Corresponding author. Tel.: +60 45995867; fax: +60 45941036.E-mail addresses: redac02@eng.usm.my (A. Ab Ghani),

redacazamath@eng.usm.my, mdazmath@gmail.com (H.Md. Azamathulla),celau@eng.usm.my (T.L. Lau), ravikanth0712@gmail.com (C.H. Ravikanth),redac01@eng.usm.my (N.A. Zakaria), redac21@eng.usm.my (C.S. Leow),lan_bullare2@yahoo.com (M.A.M. Yusof).

0955-5986/$ – see front matter© 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.flowmeasinst.2011.02.004

plastic bags, polystyrene, papers, cans, twigs, leaves, boxes andold clothes. A study conducted in Melbourne, Australia notedthat urban areas contribute about 20–40 kg (dry mass) of grosspollutants per hectare per year to stormwater, equivalent toapproximately 60,000 tons or 230,000 cubic meters of grosspollutants and about two billion items of litter annually [1].

The increase in urban population density and built-up areas di-rectly or indirectly affects hydrological processes through changesin stormwater runoff or the stream flow regime, alterations inpeak flow characteristics, decrease in water quality, and changesin a river’s amenities. A shift in land use from agriculture tourban development generally removes vegetated land cover andcontributes to increased surface imperviousness. This increasedimperviousness will lead to more erosion of land surfaces. Eventu-ally, the eroded sediment will be transported and then depositedin a waterway. Industrialization and urbanization are the mainhuman activities that affect the environment by increasing themagnitude of pollution, thus, increasing the amount of sedimentdeposited in conventional drainage systems in Malaysia [2]. Mostof the sediment deposition in conventional drainage systems iscaused by human activities, such as construction and road work;some deposition is also due to erosion from the surrounding areas.Gross pollutants and sediment deposits decrease the ability of ur-ban drainage systems to convey stormwater runoff. As a result, ur-banization increases flood runoff and can produce hazardous flash

216 A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224

Fig. 1. Various types of gross pollutants in an industrial area, the Mak-Mandin Industrial park in Penang.

flooding. A detailed study on the sediment characteristics in a con-ventional drainage system is an essential process to help overcomeand reduce the incidence of problems related to stormwater man-agement, especially sediment deposition in drains.

Land use changes due to urbanization had massive effectson the biological integrity of the receiving waters as rainfall-infiltration decreases and stormwater runoff volume increases [3].Ab. Ghani et al. [4] carried out a preliminary study on severalsites in the northern region of Peninsular Malaysia to investigatethe types of pollutants and sediment characteristics in thestormwater system. Photographs were taken to classify the typesof gross pollutants according to the site location. The sedimentsamples also were collected at each station using a grab samplingtechnique and were analyzed in the laboratory to establish thesediment size distribution using dry sieve analysis. Sedimentsampling at five different locations in the same study revealedthat the mean sediment size of the deposited sediment variedfrom 0.6 mm to 0.74 mm. The study also noted that sedimentwas mainly transported as bed load and was classified as non-cohesive sediment. Bed load refers to sediment particles that aretransported in a thin layer of the order of two grain diametersthick next to the bed. Fig. 2 shows the sediment size distributionof samples from the five sampling locations. Ab. Ghani et al. [4]reported that the average particle size for sediment deposits foundin the urban drainage systems of five cities in Malaysia (JohorBahru, Kota Bharu, Alor Setar, Ipoh, and Butterworth) ranged from0.6 to 0.9 mm and were mostly non-cohesive. Ninety percentof the samples consisted of sand and gravel. The same studyalso stated that the major sources of sediments in the drainagesystems for these five cities are construction activities, erosionfrom surrounding areas, and road work.

The sediment and gross pollutants in the storm drains haveresulted in increased overbank flooding resulting from the lossof hydraulic capacity of the drains. The Department of Irrigationand Drainage (DID) in Malaysia recommends a constant minimumvelocity of 0.9 m/s to minimize sedimentation problems [5]. Theself-cleansing approach helps minimize the sediment depositionproblems in drainage systems. Ab. Ghani et al. [6] applied a designof a sewage and drainage system for self-cleaning purposes andnoted that the range of theminimumvelocitywas between 0.6m/sand 1.5 m/s.

Fig. 2. Mean sediment size distribution in five locations of the Northern region ofMalaysia.

2. Gross Pollutant Traps (GPTs)

Gross Pollutant Traps are one of the components that havethe ability to treat stormwater and reduce the flow energythrough their ‘self-cleansing ability’ [2]. If the screens are blocked,stormwater is still able to pass through the structure withlimited upstream flooding. As a result, screen-less or limitedscreen traps are not advisable for many applications [7]. Thereduction and removal of urban litter is a complex and difficultproblem, particularly for developing countries. Ultimately, thesolution depends on each local authority developing an integratedcatchment litter management strategy that includes planningcontrols, source controls, and structural controls [8]. GPTs inBrookvale Creek, in Sydney’s Northern Beaches region, use theoutlet approach inwhich the GPTs act as the single treatment pointfor the upstream catchments [9]. This type of GPT is installed in-line and functions via direct screening with a grid or mesh barrierassembly to reduce the quantity of gross pollutants carried by therunoff. The GPT traps the coarse sediment by decreasing the flowvelocity. The GPT that was used in this creek hasmarkedly reducedthe amount of gross pollutant in the stream [10]. Recent studies ofGPTs in Brisbane, Australia indicate that weather conditions suchas the extent and duration of rainfall can influence the inflow rate

A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224 217

Secondary TrashRack

Secondary channel

Sediment Trap

Water CollectionTrap

Primary TrashRack

Valve to maintainuniform flow

Fig. 3. 3-dimensional view of a GPT prototype.

Fig. 4a. Plan view of a Gross Pollutant Trap.

to theGPTs. GPT blockage, either partial or full, can change the litterretention characteristics of the structures ([11] as cited in [12]).

3. Experimental procedure

A Gross Pollutant Trap (GPT) is an engineered sediment trapdesigned to trap and remove litter, debris, and coarse sedimentfrom runoff. The main function of GPT is to keep coarse sedimentout of stormwater systems and to protect vegetation and habitatsin natural bodies of water from the smothering effects of sediment.GPTs may also be used in the pretreatment of flow into a pondor wetland to confine the area of deposition of coarse sediments.With proper modifications, GPTs have been documented to alsoprovide some reduction in other pollutants, such as particulatenutrients, trace metals, oil and grease, bacteria, and dissolvedoxygen-demanding substances. In this study, the proposed GPT

was designed to trap sediment and gross pollutants of stormwaterrunoff from roofs, yards, roads or lawns in an urban area.Additionally, designs of the GPT also include consideration forpollutant separation by the energy of thewater flow (self-cleansingmethod).

The GPT used in this study was a multi-component systemconsisting of a silt trap and Gross Pollutant Trap capable ofpreventing bed load and removing solid waste from stormwater.Fig. 3 shows the structure of the GPT. Basically, it consisted oftwo compartments: a primary trap and a secondary trap. Theprimary trap was a sediment trap compartment equipped with aprimary trash rack,whichwas constructed in-linewith the channelflow direction to treat stormwater during low flow conditions.The primary trap comprised a uniform channel with an expansionextended from the existing drain and with a drop at the sedimenttrap to reduce the velocity of the incoming flow. It is essential

218 A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224

Fig. 4b. Sectional view of a GPT.

Fig. 4c. Sectional view of a GPT prototype. Y = 400 mm, θ = 31°.

Fig. 4d. Sectional view of a GPT prototype.

A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224 219

Fig. 5. Experimental setup for the Gross Pollutant Trap in the REDAC lab, USM.

Table 1Hydrological data.

Location Penang

Catchment area, A 50 000 m2

Time of concentration, tc 30 min

Description Design recurrent interval3-month ARI 5-year ARI 100-year ARI

Rainfall intensity (mm/h) 49.42 116.51 186.35Runoff coefficient 0.79 0.87 0.895Discharge (m3/s) 0.55 1.41 2.32

to reduce the flow velocity to achieve optimal settlement of thesediment. Under low flow conditions, turbulence is expected tobe less significant. Denser pollutants will settle out of the watercolumn and onto the bottom of the sediment trap, while otherfloatables will be trapped at the primary trash rack; the water willcontinue to filter through the trap. The primary trap was designedto serve a catchment area up to a 3-month average recurrentinterval (ARI) event, a design requirement set for stormwaterquality treatment in Malaysia.

The secondary traps are present on both sides of the primarytrap for the purpose of treating stormwater runoff from eventslarger than 3-month ARI. In addition, the secondary traps alsofunction as back-up traps to treat stormwater runoff whenever theprimary trap is blocked. The excess water will spill sideways intothe secondary traps and subsequently into the secondary channelbefore finally rejoining the flow discharging from the main trashrack. The secondary trap compartments are designed using theself-cleaning principle. The pollutants will be intercepted by thescreen and forced down into it by a combination of themomentumof water and gravity until the pollutants come to rest in a wastecollection bin. These bins are allowed to dry out for removal of the

pollutants. A similar principle is used in the Stormwater CleaningSystem (SCS) type of GPT (Armitage and Roosboom [13]).

The dimensions for the GPT prototype are shown in Figs. 4a–4d. The GPT is to serve a catchment area of 50 hectares with a30-min concentration time. Table 1 summarizes the computedcatchment properties and design flow for various storm events.The experimental setup is presented in Fig. 5. This GPT prototypewas constructed based on the Froude Number similarity. Inpractical settings, pollutants may be found in very large range ofsizes, densities, and shapes. However, the most common grosspollutant types found inMalaysian stormwater system (debris andsediment) were used in this laboratory test. The settling velocitiesand densities of the typical gross pollutants were determined, andsimilitude laws were then used to identify representative scaleparticles. Tests were carried out for a variety of flow rates toexamine the performance of the GPT in response to both minorand major designed storm events. As shown in Table 1, the 0.3 mdepth represents dry weather flow events, 0.5 m depth representsfrequent events and 0.7 m depth represents occasional events.

Each experiment began with the release of water at the maininlet of the GPT. The direction of water flow and cross sectionsare shown in Fig. 5. Flow velocities were measured after the flowpattern had stabilized and was sustained at a depth 0.3 m. Thevelocities were measured using an electromagnetic current meter(Fig. 6) at various points on each cross section, as shown in Fig. 7.Specific amounts of gross pollutants were added at the inlet of GPTafter the flow stabilized. As water filters through the trash screen,gross pollutants are trapped by the screen at different locationsin the main channel. Then, the GPT was drained, and the trappedgross pollutants were collected, oven-dried, and weighed. The drymass of the collected gross pollutants was compared with thatof the gross pollutant sample released during the simulation todetermine the trapping efficiency. Simulations were conducted forother flow conditions, i.e., for depths of 0.5m and 0.7m in themain

Table 2Gross pollutant trapping efficiency.

Flow depth (cm) Flow discharge (l/s) Gross pollutants (kg) Pollutants trapped (kg) Trapping efficiency (%)

30 14 11 9.6 8750 46 13.75 11.5 8470 150 16.5 13.48 82

220 A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224

Fig. 6. Measuring the flow velocity using flow-rate equipment.

Fig. 7. Velocity measurement points at sections A, B, and C.

channel, and the results recorded. For the simulation at a depth of0.7 m, the secondary traps were also involved because the waterdepth exceeded the overspill crest level.

For the sediment analysis, the system was placed at the flowterminal, which was made up of a container, a sediment tray andan electromagnetic vibrator. The container was filled with drysediment and connected to the sediment tray using a funnel. Therate of supply was controlled using a controller. The container wasfilled with sand. The sand was released into the uniform flow atthe flow terminal for a fixed period of time. To maintain a constantrate of sediment supply, the amount of sand was kept even withthe top of the container. The constant sediment supply rate wasmeasured by weighing the amount of sediment collected over afixed period of 2 min. Finally, the system supplied sediment intothe flow terminal for 30 min.

4. Results and discussions

Fig. 8 illustrates the rating curves that were developed at crosssections A, B and C. The flow discharge increased in conjunctionwith the flow depth. Fig. 9 shows the flow patterns at sections A,B, and C; the flow velocity was highest at section A. Figs. 10a–10cshow the velocity profiles for various cross sections atwater depthsof 0.3 m, 0.5 m, and 0.7 m, respectively. The velocity distributionfor the right bank from the middle section is higher than the leftbank.

4.1. Gross Pollutant Trap efficiency

The trapping efficiency is generally expressed in terms of thepercentage of particles trapped, and the trapping efficiency was

0.7

0.6

0.5

0.4

0.3

Dep

th (

m)

60 70 80 90 100

Discharge (I/s)

110 120 130 140

Flow Discharge

Fig. 8. Flow rating curve for the GPT prototype.

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.50

0.00

Vel

ocity

(m

/s)

0.1 0.2 0.3 0.4 0.50.0 0.6

Depth (m)

Section A

Section B

Section C

Fig. 9. Flow pattern in the GPT.

measured for a wide range of flow rates and water depths. Table 2summarizes the trapping efficiency for gross pollutants by the GPT.The trapping efficiency for a water depth of 0.3 m was slightlyhigher than that for a depth of 0.7 m. Fig. 11(a)–(c) show theobservational evidence of the trapping efficiency for depths of0.3 m, 0.5 m, and 0.7 m, respectively.

The amount of trapped floatable gross pollutants at a waterdepth of 0.3 m was greater than those at depths of 0.5 m and0.7 m. This laboratory observation can be related to actual siteconditions, where an extreme rainstorm will generate a largevolume of stormwater that could transport gross pollutants furtherdownstream and into natural bodies ofwater. Logically, as the flowincreases, gross pollutants, especially floatables, easily flow overthe GPT and are transported downstream, making conventionaltrash screens orGPTs less effective during higher flows. The currentGPT overcomes this weakness by incorporating secondary traps,which allow higher flow to overspill through the side crest (ratherthan directly above the primary screen). The secondary traps areequipped with additional screens to filter out pollutants whenwater starts spilling over into these traps, hence, maintaining theefficiency level of the GPT, even in higher flow conditions.

4.2. Sediment trapping efficiency

The sediment trapping efficiency of the GPT was investigatedusing the average rates of sediment supply of 2.53 g/s and 3.169 g/s

A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224 221

Table 3Sediment trapping efficiency.

Flow depth (cm) Discharge (l/s) Input rate (g/s) Amount of sediment released (g) Amount of sediment trapped (g) Trapping efficiency (%)

30 41 2.68 4824 3442 71.353.18 5730 3993 69.68

50 88 2.38 4275 2889 67.573.2 5760 3653 63.42

70 161 2.53 4554 2863 62.863.13 5625 3525 62.66

Fig. 10a. Velocity plot for the 0.3 m water depth.

Fig. 10b. Velocity plot for the 0.5 m water depth.

222 A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224

Fig. 10c. Velocity plot for the 0.7 m water depth.

Fig. 11. Fig. (a), (b), and (c) show visual evidence for the trapping efficiencies at depths of 0.3 m, 0.5 m and 0.7 m, respectively.

at the main inlet of the GPT (Fig. 12). The flow diagrams inFig. 13 show the physical and observational data for sedimenttrapped in the main channel of the GPT. Table 3 presents theamount of sediment trapped in the GPT for an assumed time

of concentration (30 min). The trapping efficiency was high forlower water depths and low for higher water depths. In otherwords, the efficiency is inversely proportional to the water depthin the GPT. This relationship can be explained by the sediment

A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224 223

Fig. 12. The sediment supply apparatus at the flow terminal was made up of acontainer, a sediment tray and an electromagnetic vibrator.

transport capacity of the flow, which increases as flow ratesincrease. During higher flow, more sediment is carried or re-suspended by the water column, hence decreasing the trappingefficiency.

The GPT performedwell at a depth of 0.3 mwith the depositionoccurring over the entire basin. Most of the sediments weredeposited at the inlet of the sediment basin (section A). The coarsersediment was deposited at this section, and the finer sedimentwas transported and deposited at sections B and C. The flowvelocities gradually decreased toward the outlet of the sedimentbasin. A maximum velocity of 0.32 m/s was measured at sectionC. For the sediment supply with an average rate of 2.53 g/s,most of the sediment was deposited at the inlet of the sedimentbasin, and the maximum velocity of 0.42 m/s was measured atsection A.

5. Conclusions

The purpose of this research was to investigate the hydraulicperformance and trapping efficiency of a Gross Pollutant Trap(GPT) that was designed and patented by the River Engineeringand Drainage Research Centre (REDAC), Universiti Sains Malaysia.The investigation was carried out on a laboratory prototype modelusing flow rates and gross pollutant characteristics commonlyfound in Malaysia. Measurements of the flow pattern, velocitydistribution, amount of pollutants trapped, and the pattern ofsediment deposits were recorded and analyzed. From the testresults, the following conclusions can be drawn:

1. The gross pollutant trapping efficiency was the highest at thelowest water depth tested (0.3 m). The efficiency decreased aswater depth increased. However, the introduction of the sec-ondary traps overcame this weakness by providing a secondaryscreening mechanism, thus, minimizing the loss of efficiencyduring higher flows. Overall, the GPT maintained a trapping ef-ficiency of over 80% for the three tested flow conditions.

2. The sediment trapping efficiency was studied using coarse sed-iments. The result shows that the sediment trapping efficiencyfor lowestwater depth (0.3m)was the highest (71.35%). Incom-ing sediment was mainly deposited at the inlet of the sedimentbasin. For all tested water depths, most of the sediment wastrapped within the GPT due to the low flow velocity zone cre-ated by the expansion of the initial channel into the sedimentbasin.

3. Further studies are being planned to investigate the perfor-mance of the Gross Pollutant Trap with respect to different pol-lutant profiles and trash screen sizes.

Acknowledgement

This study was carried out under a research university grantproject ‘‘Integrated Urban Drainage Management (INUDRAM)’’(001/REDAC/814014), Universiti Sains Malaysia. The authorswish to thank Robert D. Jarrett, National Research ProgramPaleohydrology and Climate Change, US Geological Survey (USGS),for his suggestions regarding the preparation of this manuscript.

Fig. 13. Sediment trap at various depths.

224 A. Ab Ghani et al. / Flow Measurement and Instrumentation 22 (2011) 215–224

References

[1] Allison RA, Chiew FH, McMahon TA. Stormwater gross pollutants. Report97/11. Australia: Co-operative Research Centre for Catchment Hydrology,eWater CRC, University of Canberra; 1997.

[2] Ab. Ghani A, Lau TL, Zakaria NA, Abdullah R, Chang CK. Development ofengineered sediment trap for conventional drainage system. In: R & Dcolloquium on river engineering & urban drainage. 2005.

[3] Urbonas BR. Effectiveness of urban stormwater bumps in semi-arid climate.In: Conference paper. Session 1. Conference program: experienced with bestmanagement practices BMPs. 2003 [in Colorado].

[4] Ab. Ghani A, Zakaria NA, Kassim M, Ahmad Nasir B. Sediment sizecharacteristics of urban drains in Malaysian cities. Journal of Urban Water2000;2(4):335–41.

[5] Ab. Ghani A, Zakaria NA, Kassim M. Sediment deposition in a rigid monsoondrain. International Journal of River Basin Management 2008;6(1):23–30.

[6] Ab. Ghani A, Kassim M, Abdullah R, Mohd Sidek L, Zakaria NA. Sedimentdeposition in storm drains of Raja catchment, Alor Setar. Bulletin Institutionof Engineers Malaysia 2001;43–8.

[7] Armitage N, Roosboom A. The removal of litter from stormwater con-duits in developing world. Water Science and Technology 1999;39(9):277–284.

[8] Armitage N. The reduction of urban litter in the storm water drains of SouthAfrica. Urban Water Journal 2007;4(3):151–72.

[9] CSIRO . Urban stormwater: best practice environmental managementguidelines. Collingwood (VIC): CSIRO Publishing; 1999.

[10] Rawson CA, Sheechan RA, Skelton S. Effect of the pollutant trap onwater quality of brookvale creek. UTS freshwater ecology report. Sydney:Department of Environment Sciences. University of Technology; 2002.

[11] Thackston EL, Shields DF, Schroeder PR. Residence time distributionsof shallow basins. Journal of Environmental Engineering 1987;113(6):1319–1332.

[12] Madhani JT, Kelson NA, Brown RJ. An experimental and theoretical investiga-tion of flow in a gross pollutant trap. Water Science and Technology 2009;59(6):1117–27.

[13] Armitage N, Roosboom A. The removal of urban litter from stormwaterconduits and streams: Paper 3 — Selecting the most suitable trap. WaterScience and Technology 2000;26(2):196–7.